Finds photographs recursively, preserving their relative paths. With a batch/control and batch/treatment folder structure, each batch path proposes a separate set_id and the leaf folder proposes the treatment. Only explicit control folder names are recognized. Other structures produce blank labels that may remain blank for measurement; matching labels are needed for comparisons. Folder suggestions never confirm experimental identity or QC. One photograph must represent one independent plate; repeated time points of a plate are not independent replicates.
Usage
prepare_experiment(
input = NULL,
project_dir = "FunMel_project",
metadata = NULL,
control_folders = c("control", "controls"),
max_uncompressed_mb = 2048
)Arguments
- input
Photograph folder, ZIP, or a photograph inside the intended folder. Selecting a photograph includes all supported images in its containing folder and subfolders, not only that image. For multiple batches, supply the root folder or ZIP. NULL opens a file chooser. Selecting an existing project or analysis RDS beside project.rds reopens it with load_experiment().
- project_dir
New output folder; must be empty or not yet exist.
- metadata
Optional metadata data frame or CSV path. Otherwise a unique plate_metadata.csv in the input is used, or a template is proposed.
- control_folders
Exact folder labels denoting controls (case-insensitive).
- max_uncompressed_mb
Maximum total selected ZIP entry size in MB.
Examples
if (FALSE) { # \dontrun{
project <- prepare_experiment() # Choose a photo ZIP or one photo in its folder.
# Matching labels can be supplied later; use the correct segmentation layout.
result <- run_experiment(project, n_spots = 5, spot_radius_mm = 0.4)
# Inspect project$qc_dir; review reminders do not require approval-flag edits.
reviewed <- finish_experiment(project)
} # }